Odd coefficient count for an ideal above-smoothing Gaussian pool
ProvedInfoLossLAR.WorstCase.RegevSketch.dual_parity_odd_count_bounddiscrete-gaussianlar-regev-bridgelattices
At twice the smoothing parameter, the discrete Gaussian reduced modulo two is close to uniform. Thus an individual coefficient is odd with probability at least when the smoothing error is at most (using the pointwise coset bounds). For independent draws, a binomial deviation bound makes the probability of fewer than a quarter of the draws being odd at most eventually. The full countable product distribution has total mass one.
Preamble
import Definitions.Def_InfoLossLAR_RegevSketch_DGS set_option autoImplicit false set_option maxHeartbeats 200000 open QAlgorithms.Complexity open scoped BigOperators Classical
Formal statement
namespace InfoLossLAR.WorstCase.RegevSketch
theorem dual_parity_odd_count_bound :
∃ N : ℕ, ∀ n ≥ N, 1 ≤ n → ∀ (D : Matrix (Fin n) (Fin n) ℤ), D.det ≠ 0 →
∀ r : ℝ, 0 < r → smoothingError n ≤ 1/8 →
RegevLWE.GaussConv.smoothingParam (inputModule D) (smoothingError n) ≤ r/2 →
let m := (n+1)^3
HasSum (fun p : Fin m → Fin n → ℤ => ∏ j, coefficientMass D r (p j)) 1 ∧
∀ i : Fin n,
(∑' p : Fin m → Fin n → ℤ, (∏ j, coefficientMass D r (p j)) *
(@ite ℝ (¬ (m ≤ 4 * (Finset.univ.filter (fun j => Odd (p j i))).card))
(Classical.propDecidable _) 1 0)) ≤ 1/(32*((n : ℝ)+1)) := by sorry
end InfoLossLAR.WorstCase.RegevSketch
Source
Oded Regev, On Lattices, Learning with Errors, Random Linear Codes, and Cryptography, full version dated May 2, 2009, https://cims.nyu.edu/~regev/papers/qcrypto.pdf. Lemma 3.20 and its preceding reduction from GapSVP to the promise GapCVP variant, full-version PDF p. 30; fixed-family substitution and explicit uniform decision algorithm. Alternative index-two parity and rational covariance specialization of the dual certificate.